A Hybrid Machine Learning Framework for Enhancing Energy Efficiency in Cyber-Physical Cognitive Radio Sensor Networks

Cyber-Physical Systems (CPS) integrated with Cognitive Radio Sensor Networks (CRSNs) present significant opportunities for intelligent IoT applications but also faces substantial challenges in achieving optimal energy efficiency while maintaining seamless performance. The resource-limited and heterogeneous characteristics of CPS-enabled CRSNs creates a complex computational and communication for real-time 6G applications. To address these challenges, we propose a novel hybrid machine learning framework integrating Random Forest (RF) and Neural Networks (NN) specifically designed for CPS-oriented CRSNs. The hybrid model improves classification accuracy and energy optimization by incorporating RF predictions as additional input features for NN by leveraging the complementary strengths of both techniques. The proposed framework is evaluated and compared with existing machine learning approaches including Support Vector Machines (SVM), Decision Trees (DT), and K-Nearest Neighbors (KNN) in simulated CPS environments using synthetic datasets. The proposed work’s results demonstrate that the hybrid model consistently outperforms traditional methods in terms of accuracy, energy consumption, and classification reliability, validating its effectiveness for energy-efficient communication in CPS-driven CRSN applications across diverse IoT scenarios.